Commons:Structured data/Computer-aided tagging/Blocklist
Request new items on the talk page.
Purpose of the blocklist
editThe Suggested Tags feature receives suggestions from a machine learning system built by Google. Some of those suggestions are not quite suitable for Commons for various reasons, so the teams developing the tool implemented a blocklist feature that excludes them when they are suggested.
There is currently a known issue where tags for some files don’t get updated when the blocklist does. We’re working on a solution for that, but most new uploads will not be affected.
The blocklist feature exists behind the scenes, so for now items can be added to the blocklist by developers. Editors can request blocklist additions on this talk page.
Blocklist
editThe canonical list actually lives in InitialiseSettings.php, so this list may be out of date.
Gender
edit- 'Q467', // woman
- 'Q3010', // boy
- 'Q3031', // girl
- 'Q8441', // man
- 'Q1378024', //lady
Colors
edit- 'Q1088', // blue
- 'Q23445', // black
- 'Q3142', // red
- 'Q3133', // green
- 'Q943', // yellow
- 'Q39338', // orange
- 'Q428124', // violet
- 'Q3257809', // purple
- 'Q23444', // white
- 'Q42519', // gray
- 'Q47071', // brown
- 'Q843607', // beige
- 'Q10296772', // carmine
- 'Q373058', // Azure
- 'Q317802', // silver
- 'Q1670336', // tan
- 'Q2778382', // bronze
- 'Q208045', // gold
- 'Q1078214', // teal
- 'Q906936', // lime
- 'Q543923', // rust
- 'Q2015138', // salmon
- 'Q2268159', // scarlet
- 'Q767608', // sepia
- 'Q3518540', // terra cotta
- 'Q5960345', // turquoise
- 'Q3014419', // wine
- 'Q454847', //amaranth
- 'Q679355', //amber
- 'Q1324818', //apricot
- 'Q372973', //aquamarine
- 'Q797446', //burgundy
- 'Q5043987', //carnelian
- 'Q2541418', //celadon
- 'Q313120', //cerulean
- 'Q3309916', //chocolate
- 'Q2936397', //cinnamon
- 'Q2411228', //coral
- 'Q2730433', //cream
- 'Q303826', //crimson
- 'Q180778', //cyan
- 'Q5005364', //fuchsia
- 'Q5967009', //indigo
- 'Q650770', //khaki
- 'Q2720565', //lemon
- 'Q2294993', //lilac
- 'Q3276756', //magenta
- 'Q25203611', //Mango
- 'Q25393814', //maroon
- 'Q864152', //olive
- 'Q533047', //rose
Adjectives
edit- 'Q48997611', // vintage
- 'Q10770146', // monochrome
- 'Q6453656', // monochrome
- 'Q838368', // black-and-white
- 'Q296001', // close-up
Odd nouns
edit- 'Q1301433', // land vehicle
- 'Q1802779', // land vehicle
- 'Q6478447', // properties of water
- 'Q2083958', // pattern
- 'Q738168', // pattern
- 'Q696160', // psychedelic art
- 'Q7239', // organism
- 'Q638', // music
- 'Q11634', // art of sculpture
- 'Q11629', // art of painting
- 'Q16502', // world
- 'Q1792644', // art style
- 'Q1190554', // occurrence
- 'Q295469', // ecoregion
- 'Q1049799', // water resource
- 'Q334166', // mode of transport
- 'Q42889', // vehicle
- 'Q82821', // tradition
- 'Q3248864', // Terrestrial animal
- 'Q53875', // parallelism
- 'Q37073', // pop music
- 'Q101998', // biome
- 'Q2262382', // masai lion
- 'Q721221', // serpent
- 'Q1634416', // stock photography
- 'Q1395149', // demonstration
- 'Q12554', // Middle Ages
- 'Q83180', // roof
- 'Q167510', // bitumen
- 'Q43619', // natural environment
- 'Q309', // history
- 'Q11016', // technology
- 'Q1172903', // loch
- 'Q165848', // wind wave
- 'Q211778', //Lake District
- 'Q43238', //Poaceae
- 'Q5135744', //religious institute
- 'Q486972', //human settlement
- 'Q826939', //canard
Time-oriented terms
edit- 'Q1187312', // yesterday
- 'Q988377', // day before yesterday
- 'Q1036448', // day after tomorrow
- 'Q1209716', // tomorrow
- 'Q3151690', // today
Implementation
editThis blocklist is currently implemented via the $wgMachineVisionWikidataIdBlacklist
config variable in InitialiseSettings.php
. The code that handles the actual filtering lives in GoogleCloudVisionClient::filterIdBlocklist()
in the MachineVision extension.
WithHoldImageList
editIn addition to the blocklist, as part of a temporary hold on images of people, the system utilizes a separate list of identifiers that indicate a person is in the image and the image should be excluded from the suggestion queue. This list is currently implemented via the $wgMachineVisionWithholdImageList
config variable in InitialiseSettings.php
.
- 'Q467', // woman
- 'Q3010', // boy
- 'Q3031', // girl
- 'Q8441', // man
- 'Q1378024', // lady
- 'Q255274', // white collar worker
- 'Q327968', // facial expression
- 'Q41055', // forehead
- 'Q2472587', // people
- 'Q1155908', // elder
- 'Q15173', // lip
- 'Q82714', // chin
- 'Q1886338', // makeover
- 'Q3080415', // Jheri curl
- 'Q170579', // laughter
- 'Q28472', // hair
- 'Q327496', // hairstyle
- 'Q371174', // gesture
- 'Q82714', // chin
- 'Q9633', // neck
- 'Q1886338', // Makeover
- 'Q1922956', // black hair
- 'Q1255864', // fun
- 'Q14130', // long hair
- 'Q202466', // blond hair
- 'Q327496', // hairstyle
- 'Q37017', // face
- 'Q43748', // eyebrow
- 'Q1190554', // occurrence
- 'Q82821', // tradition
- 'Q182832', // concert
- 'Q132241', // festival
- 'Q349', // sport
- 'Q2755547', // individual sport
- 'Q874405', // social group
- 'Q327245', // team
- 'Q12068677', // selfie
- 'Q23640', // head
- 'Q83360', // thumb
- 'Q36864', // nail
- 'Q33767', // hand
- 'Q319604', // passenger
- 'Q205398', // social work
- 'Q7242', // beauty
- 'Q12684', // fashion
- 'Q749212', // gentleman
- 'Q639669', // musician
- 'Q184485', // performing arts
- 'Q855091', // guitarist
- 'Q159992', // surfing
- 'Q2021379', // wakesurfing
- 'Q911069', // boardsport
- 'Q8037570', // wrangler
- 'Q273283', // beggar
- 'Q134307', // portrait